RICHARDS AND BAXTER. ATOMIC WEIGHT OF COBALT, 353 



essentially the same method as before, but was then further treated by 

 six additional recrystallizations as the purpureo-chloride, with the help 

 of very pure redistilled ammonia and hydrochloric acid. The oxides 

 obtained from these preparations were finally reduced by means of pure 

 hydrogen, instead of by ammonia as before. 



Preparation of Bromine. — Here also the process of purification did 

 not differ from that previously employed. The purity of the bromine is 

 sufficiently proven by two analyses in which a known weight of silver 

 was precipitated by a slight excess of ammonic bromide made from the 

 halo£:en. 



The balance and weights used in this work were the same as those 

 described in our previous paper. The weights were carefully restan- 

 dardized, the values differing from those previously found by only a few 

 hundredths of a milligram. Since the balance was wholly free from 

 iron, no inaccuracies could have arisen from magnetic attraction, either in 

 this work or in that upon nickel. 



Owing to the fact that the specific gravity of cobalt is almost identical 

 with that of brass, no correction was required to reduce the weight of 

 the cobalt to a vacuum standard. 



The correction of +.00010 gram per gram of cobaltous bromide was 

 applied in each case to the weighings of that material. All weighings 

 were made by substitution, as usual. 



We are indebted to the Cyrus M. Warren Fund for Chemical Research 

 in Harvard University for some of our more expensive pieces of 

 apparatus. 



Method of Analysis. 



By means of the glass apparatus described in our earlier paper, cobalt- 

 ous bromide, contained in a platinum boat, was dried in a current of pure 

 dry nitrogen and hydrobromic acid gas in a hard glass tube heated to 

 about 400° ; and after the tube had been thoroughly swept out with 

 nitrogen and then by dry air, the boat was transferred to a weighing 

 bottle in this safe atmosphere. The bottle was then weighed, and the 

 VOL. XXXIV. — 23 



